Re: [emacs-bidi] Where do I start?

[Behdad Esfahbod]

Hi,

So sorry for commenting so late, but the point here is that character 
types in Unicode Reference Implementation differs from the unicode 
charater types.

In Unicode, '+-/*' has types ET,ET,ES,ON, but in reference
implementation, (that I call the encoding CapRTL -- and is used in
fribidi when you set '--charset CapRTL' or '--caprtl') these has types
ET,ON,ES,ON, infact the type for '-' differs.

On Thu, 15 Nov 2001, Alex Schroeder wrote:

> "Eli Zaretskii" <address@hidden> writes:
> 
> >   http://crl.NMSU.Edu/~mleisher/ucdata.html
> 
> I am looking at the following problem right now:
> 
> (bidi-logical-to-visual "SOLVE 1*5 1-5 1/5 1+5")
> 
>   => "1+5 1/5 1-5 5*1 EVLOS"
> 
> I don't understand why the Unicode Reference Code is giving a
> different result.  Note that FriBidi returns the same result as my
> implementation.
> 
> |Test|   Source String    | Unicode Reference  |   FriBidi 0.1.9    |
> |    |                    |        Code        |                    |
> +----+--------------------+--------------------+--------------------+
> |  14|SOLVE 1*5 1-5 1/5   |     5+1 5/1 5-1 5*1|     1+5 1/5 1-5 5*1|
> |    |1+5                 |               EVLOS|               EVLOS|
> 
> It seems that / - + are treated the same as *.  How can that be?
> 
> (bidi-get-mnemonics
>  (bidi-get-types "*/+-"))
> 
>   => (ON ES ET ET)
> 
> According to W4 and W5 / + and - should turn into EN when present
> between or adjacent to EN as in the test string.  Any opinions?
> 
> Here are the bidi types before and after resolving weak types:
> 
> (bidi-get-mnemonics
>  (bidi-get-types "SOLVE 1*5 1-5 1/5 1+5"))
> 
>   => (R R R R R WS EN ON EN WS EN ET EN WS EN ES EN WS EN ET EN)
> 
> (bidi-get-mnemonics
>  (bidi-resolve-weak-types
>   (bidi-get-types "SOLVE 1*5 1-5 1/5 1+5")))
> 
>   => (R R R R R R EN R EN R EN EN EN R EN EN EN R EN EN EN)
>                      *         *          *          *
> 
> Alex.
> 

-- 
Behdad
29 Aban 1380, 2001 Nov 20

[Finger for Geek Code]